PCRAM devices store binary data as two different resistance values, one higher than the other. The resistance value represents a particular binary value of logic “0” or logic “1”. When sensing the resistance value of a PCRAM device, it is common to compare the resistance of a memory cell undergoing a read operation with resistance of a reference cell to determine the resistance value of the cell being read and thus its logic state. Such an approach is disclosed in U.S. Pat. No. 5,883,827. However, this approach has some limitations.
If the reference cell is defective and a column of memory cells within an array uses a same defective reference cell, the entire column of memory cells will have erroneous resistance readings. In addition, specialized circuitry is required to write a resistance value into the reference cell, and a sense amplifier circuit for such an arrangement tends to be complex and large.
Typically, sensing schemes for PCRAM devices also tend to have a unique architecture which is different from that normally employed in typical DRAM circuits. Although PCRAM's differ from DRAM's in that they store binary values in resistive memory elements rather than as charges on capacitors, and although PCRAM's are non-volatile, where the capacitor structures employed in DRAM's are volatile, nevertheless it would be desirable if the read and write circuits for both devices were as similar as possible so that existing DRAM memory device architectures could be easily adapted to read and write PCRAM devices.